Abstract:We present a modified limited memory BFGS (L-BFGS) method that converges globally and linearly for nonconvex objective functions. Its distinguishing feature is that it turns into L-BFGS if the iterates cluster at a point near which the objective is strongly convex with Lipschitz gradients, thereby inheriting the outstanding effectiveness of the classical method. These strong convergence guarantees are enabled by a novel form of cautious updating, where, among others, it is decided anew in each iteration which of the stored pairs are used for updating and which ones are skipped. In particular, this yields the first modification of cautious updating for which all cluster points are stationary while the spectrum of the L-BFGS operator is not permanently restricted, and this holds without Lipschitz continuity of the gradient. In fact, for Wolfe-Powell line searches we show that continuity of the gradient is sufficient for global convergence, which extends to other descent methods. Since we allow the memory size to be zero in the globalized L-BFGS method, we also obtain a new globalization of the Barzilai-Borwein spectral gradient (BB) method. The convergence analysis is developed in Hilbert space under comparably weak assumptions and covers Armijo and Wolfe-Powell line searches. We illustrate the theoretical findings with numerical experiments. The experiments indicate that if one of the parameters of the cautious updating is chosen sufficiently small, then the modified method agrees entirely with L-BFGS/BB. We also discuss this in the theoretical part. An implementation of the new method is available on arXiv.

A Nonmonotone Filter Barzilai-Borwein Method for Optimization

The convergence of conjugate gradient method with nonmonotone line search

An Efficient Interior-Point Algorithm with New Non-Monotone Line Search Filter Method for Nonlinear Constrained Programming

Barzilai and Borwein conjugate gradient method equipped with a non-monotone line search technique and its application on non-negative matrix factorization

A global Barzilai and Borwein's gradient normalization descent method for multiobjective optimization

On the acceleration of the Barzilai-Borwein method

A globalization of L-BFGS and the Barzilai-Borwein method for nonconvex unconstrained optimization

A nonmonotone spectral projected gradient method for large-scale topology optimization problems

Analysis and Performance of the Barzilai-Borwein Step-Size Rules for Optimization Problems in Hilbert Spaces

AdaBB: Adaptive Barzilai-Borwein Method for Convex Optimization

Variational Barzilai-Borwein Method

A Tabu Search Based Metaheuristic for Fast Global Optimizations of Inverse Problems

Safeguarding adaptive methods: global convergence of Barzilai-Borwein and other stepsize choices

Stabilized Barzilai-Borwein method

A Subspace Minimization Barzilai-Borwein Method for Multiobjective Optimization Problems

Globalization of convergence of the constrained piecewise Levenberg–Marquardt method

A modified Polak-Ribiere-Polyak type conjugate gradient method with two stepsize strategies for vector optimization

On the convergence and mesh-independent property of the Barzilai–Borwein method for PDE-constrained optimization

A Nonmonotone Filter SQP Method: Local Convergence and Numerical Results

Global Convergence of the Polak-Ribière and Hestenes-Stiefel Conjugate Gradient Methods for the Unconstrained Nonlinear Optimization

On the convergence of proximal gradient methods for convex simple bilevel optimization